COPULA.RPF—Copula estimation

[TomDoan]

copula.rpf demonstrates the use of copulas as an alternative to multivariate GARCH models. It uses the data set from the garchmv.rpf example (though just a pair of series rather than all three).

Detailed description

[abi]

Hi Tom,

May i use copula method to measuring volatility?

Thanks,

[TomDoan]

I'm not sure what you mean. This uses univariate GARCH models to get the "volatility" and uses the Copula to model the interactions (the "off-diagonals").

[abi]

I'm not sure what you mean. This uses univariate GARCH models to get the "volatility" and uses the Copula to model the interactions (the "off-diagonals").

Thank's for reply,
Actually some paper using CPULA-GARCH method to measuring value at risk (VaR) and my question comes from there. in one of them the authors have used GJR version of GARCH based on student's t-copula to estimate VaR.

[TomDoan]

OK, so it's the VaR of what type of portfolio? A Copula doesn't apply to just a single asset.

[abi]

OK, so it's the VaR of what type of portfolio? A Copula doesn't apply to just a single asset.

VaR of stock market of four countries.

[TomDoan]

That doesn't answer the question. The VaR is computed for a position (or trading strategy).

[abi]

Thank's again Tom,
I don't know however the article is available at the below link
https://scholar.google.com/scholar?q=Me ... i=scholart

[PeterF]

As Tom already stated, a copula requires at least two instruments. The advantage of the copula concept is that one can combine different distributions of the instruments to one joint distribution. Thus, one might combine for example a Gaussian normal with a student t-distribution. This feature makes it attractive to apply copulas for the VaR of a portfolio with instruments having different distributions.

Best regards
PeterF

[TomDoan]

Thank's again Tom,
I don't know however the article is available at the below link
https://scholar.google.com/scholar?q=Me ... i=scholart

I would suggest starting with a different paper---check their literature review for something earlier and in a higher quality journal. They leave a lot of details out of what they're doing.

[hungufm]

Hi Tom,

Do you have the sample paper of garchmv.rpf using copula GARCH? I see the codes but do not have any info related to this example.

I am looking forward to hearing from you,

Ngo

[hungufm]

Hi Tom, please tell me parts of user's guide 10 where I can find info related to the example COPULA.RPF.
Thanks.

[TomDoan]

It's not covered in the manual and copula.rpf is just an example of the calculations---it's not taken from the literature.

[hungufm]

Hi Tom,

There are two kinds of copula models. First is constant and second is time-varying copula model.

May i ask you that we can estimate both models in RATS 10?

Please give me advice,

[TomDoan]

The example is for constant copula.